3.2.1

Modelling and Simulations

A finite element analysis was performed on GMOD in order to verify that the structure of the payload can withstand the loads experienced during launch of the satellite. The EIRSAT-1 Computer Aided Design (CAD) model was used to develop a Finite Element (FE) model of the structure with Autodesk Inventor and Autodesk NASTRAN-in-CAD.³⁷ The GMOD model was constrained by the PCB corner holes representing the mounting configuration while integrated within in the EIRSAT-1 stack. Normal Mode Analysis was performed on the GMOD FE model to calculate the natural frequencies and mode shapes of GMOD. For each mode, the modal effective masses in each Degree of Freedom (DOF) was determined, with modes >10% being used to inform the placement of the accelerometers for the vibration campaign. A number of loading cases were then simulated in all three axes, including quasi-static acceleration, sine vibration and random vibration. These were used to determine the Margin of Safety (MoS) with respect to the yield stress calculated using the Von Mises stress in each material for a given loading direction. The MoS was calculated as the ratio between the design allowable stress and the design limit stress scaled by a factor of safety of 1.25 per ECSS-E-ST-32-10C (2009).³⁸ Simulations show that while positive, the minimum calculated MoS (throughout the spacecraft) of 3.1, 22.4 and 2.7 occurred in the GMOD MB along the Z-axis direction during quasi-static, sine vibration and random vibration loading respectively. This reflects the effect of having a large concentrated mass such as the GMOD detector assembly mounted atop the MB PCB. The calculated MoS for the MB is shown in Table 1 for all three axes and loading cases.

Table 1:

Quasi-static, sine and random vibration Von Mises stress and MoS calculations for the GMOD MB PCB. Note: for the random vibration values the V.M stress values are the 3σ measurements.

3.2.2

Test Levels and Set-Up

Prior to any testing at the CSF, GMOD and accompanying EGSE were unpacked and cleaned with Isopropyl Alcohol (IPA). A visual inspection and functional check was then performed to ensure that no damage had occurred during transport. Three measurement accelerometers were installed on GMOD by the CSF operators. The position of the accelerometers were chosen to detect the primary modes during vibration as determined from the FE analysis described in Section 3.2.1. Accelerometers A1 and A2 were installed on the +Z face of the detector housing and the GMOD MB surface respectively (Figure 5A), while accelerometer A3 was placed on the -Z face of the MB surface (Figure 5B). GMOD was mounted to the shaker table via two custom machined aluminium mounting plates: the Shaker Interface Plate (SIP), which was mounted to the shaker, and the Subsystem Adaptor Plate (SAP), which mounted on the SIP and which interfaced GMOD to the shaker table (Figure 5C). Control accelerometers C1 and C2 were mounted on the SIP as close as possible to the mounting position of the instrument. GMOD was attached to the SIP via the four mounting holes on the MB. To replicate the flight configuration of the payload in the EIRSAT-1 PCB stack, 5mm PTFE spacers centred on the MB mounting holes were placed above and below the MB PCB. The mounting screws for the SAP and GMOD were marked with permanent marker during installation so any rotation could be observed after vibration. The spacers and screw markings can be seen in Figure 5 at each corner of the MB PCB. The levels and durations for the sine³² and random³⁹ vibration tests, including tolerances, are detailed in Table 2. As the launch vehicle is yet to be determined (International Space Station (ISS) orbit presumed), the vibration test levels are selected based on qualifying the payload for a variety of launch providers such that they cover the minimum requirements for the JAXA H-II Transfer Vehicle (HTV), SpaceX Dragon or Northrop Grumman Orbital Cygnus spacecraft.^{40, 41}

3.2.3

Vibration Test Operation & Results

Sine and random vibration tests were performed for each axis with respect to the instrument’s CRF. Sine vibration was conducted as a sinusoidal sweep from 5 up to 100Hz at 2 octaves per minute. The random vibration sequence was comprised of consecutive 10s intervals relative to the specified test levels and then a final interval of 120s at 0 dB. Resonance searches, consisting of a low level sine sweep, were performed before and after each sine and random vibration run. This was used to generate a frequency response of the structure which can be used as comparison pre/post vibration to identify any shift in resonance amplitude or frequency revealing structural changes, including non visible damage, which may have occurred during the test. Visual inspections of the instrument and the screw alignment marks were performed after each sine and random vibration. A functional check was initially scheduled to be performed after completion of the vibration testing, however this was later amended to be performed after completing each axis vibration, to verify that the payload survived without sustaining damage. The functional test involved a basic health check of the instrument which ensured all SiPM channels were still functional; any missing channels would suggest damage to either the SiPM array, or the sensitive wire bonded connections on the CoB SIPHRA. Once the health check was successfully completed, a spectrum was generated using the 2% thoriated welding rods which allowed a rough gauge of the instrument performance; any significant variation in resolution or gain might indicate crystal decoupling as some form of light (signal) loss in the transmission between the scintillator and the SiPM array. The test was considered passed once GMOD successfully produced a spectrum with distinguishable lines from the thoriated welding rods.

Table 2:

Vibration test levels and durations for sine, random and resonance searches

To pass the vibration test campaign, a series of pass/fail criteria were defined. Those most relevant are listed below. The vibration test for each axis would be passed if:

Figure 5:

GMOD mounted on the shaker for Z-axis vibration. One of the SIP control accelerometers is visible just beneath the MB PCB. A) Location of the accelerometers A1 and A2. B) Location of accelerometer A3 on the underside of the MB PCB. C) The mounting and orientation of GMOD on the SAP and interface to the shaker using the SIP.

Figure 6 shows the pre- and post-resonance search plots for the X-axis vibration. Each column represents resonance data from a single accelerometer (A1, A2, A3) mounted on GMOD, with each row showing acceleration measurements in the X, Y and Z-axis directions with respect to the CRF. Finally, as specified by the legend, each cell is populated by resonance search profiles for pre-sine, post-sine and post-random (RS1, RS2, RS3) tests, relative to the given accelerometer and axis.

While all other tests passed the pass/fail criteria, the X-axis random vibration failed due to a percentage peak shift greater than the predefined limits in amplitude and frequency. The X-axis was the first axis to undergo vibration testing, starting with a pre-sine resonance search, followed by the sine vibration and post-sine resonance search. Sine vibration was completed successfully, seen in Figure 6 as the RS1 (blue) and RS2 (red) profiles overlap nearly perfectly with little shift in frequency or amplitude. Following this, a visual inspection was performed with no noticeable change to the instrument or its mounting on the shaker table. Random vibration and post-random resonance searches were then conducted. The post-random resonance search profile, RS3 (grey), shows a clear shift in amplitude and frequency after random vibration, as seen in Figure 6. For example, a frequency shift of 11.5% was observed for the lowest mode, which violates the first pass/fail criteria stated above. Resonance frequency shifts based on the most prominent resonance peaks for both sine and random vibration tests in the X-axis are shown in Table 3 and may be compared to the resonance profiles in Figure 6.

Table 3:

The percentage frequency and amplitude shifts recorded for resonance searches pre/post sine vibration and pre/post random vibration for all accelerometer axes during X-Axis vibration. Any entries highlighted violated the pass/fail criteria. These shifts have been measured from the most prominent resonance peak in both sine and random vibration tests. All other axes passed with shifts less than the predefined limits.

Figure 6:

X, Y and Z-axis direction resonances (rows) measured during the X-axis vibration testing for each accelerometer (columns) placed on GMOD. Pre/post sine (RS1–blue and RS2–red) and post-random (RS3–grey) resonance searches profiles are shown in each cell whose percentage shifts may be referenced in Table 3.

Figure 7:

Spectrum generated by GMOD using the thoriated welding rods post X-axis vibration indicating no loss in performance. The expected lines (as referenced in Figure 4) are observable and have been highlighted. The low number of counts is a result of the short integration time being ~10 minutes in duration. The spectrum has been re-binned from single bins (ADC channels) to 8 channel wide bins to help distinguish the peaks.

A visual inspection of the GMOD MB and detector assembly exterior followed, with particular attention paid to PTFE spacers supporting the MB for any shifting or rotation about their centres which may have occurred. No visual changes were noticed and none of the marked spacers indicated any such movement. Due to this shift in frequency a functional test was performed after X-axis vibration. GMOD successfully produced a gamma-ray spectrum with evident spectral lines for the ²³²Th source during the functional test, as shown in Figure 7. As GMOD was fully functional, it was deemed appropriate to continue with the Y-axis and Z-axis vibration tests. In the original test procedure, functional testing was foreseen only at the end of the vibration test campaign. After the X-axis anomaly was identified the procedure was modified so that a visual inspection and reduced functional test was carried out after vibration of each axis to monitor the functionality of GMOD throughout the campaign.

Between axes vibration, GMOD was dismounted from the SAP. Between the Y-axis and Z-axis set up, an opportunity to inspect the underside of the MB became available; there was no indication of damage found. After vibration in the remaining axes, all pass/fail criteria were passed successfully, no anomalies were detected and GMOD was proven to be fully functional as a γ-ray spectrum with clear spectral lines was produced in the functional check. GMOD was then removed from the shaker table and disassembled for the mounting of thermocouple sensors for the TVAC testing. This gave a valuable opportunity to visually inspect the electronics assembly hardware post vibration testing to locate any internal structural changes. Each component in the electronics assembly was carefully disassembled, photographed and cleaned using IPA and lint free wipes. No damage was observed. The following are the potential causes of the frequency shift observed during the X-axis vibration which were derived after root-cause analysis, ordered from most to least likely.

None of these proposed causes pose any issues for GMOD should they occur during the mission. The vibration test campaign was completed successfully and provided confidence in the mechanical design of GMOD. A NCR was issued for the X-axis shift, and the environmental testing continued as planned.

3.3.1

Pre-Environmental Bake-Out

Before being shipped to the CSF, the finished GMOD EQM components were baked out under vacuum to prevent the outgassing of contaminants within the CSF TVAC. During thermal-vacuum testing, neither a Residual Gas Analyzer nor a Thermoelectric Quartz Crystal Microbalance were used to monitor outgassing, however a cold trap was used to collect any contaminant which may not have already been outgassed during bakeout. Additionally all other equipment (harnessing, adapters etc.) to be used within the components were verified to be vacuum compatible⁴² before transport.

3.3.2

Modelling and Simulations

Simulations were used to determine the expected temperature profile of the satellite while in orbit. A thermal analysis of the spacecraft was performed using the C&RTech package, Thermal Desktop*, an AutoCAD plugin for general thermal response modelling. The spacecraft model was simplified into a set of geometric bodies with curved surfaces approximated by polygonal structures of similar cross-sectional area. This was done to accelerate the computation time. Simplifications include rounded corners being squared, holes being removed from PCBs, the SiPMs being substituted by silicon wafers and bolts being removed. For other simple volume components (like resistors, capacitors etc), these were assimilated into a single lump mass across the PCB surface. The MSP430 and CoolRunnerII CPLD were retained for their thermal contribution as their profile would vary for operational and non-operational modes. The CeBr₃ crystal was modelled as a single block of Cerium for a conservative estimate of the thermal properties, as at the time this model was developed CeBr₃ had only just began to be used for scintillation and few sources on the thermal properties were in existence. It was anticipated that the crystal would be heavily insulated due to being wrapped in a number of layers of teflon (as discussed in Section 2) and hermetically encased within its own aluminium housing. The satellite model was simulated in an ISS-like LEO with an initial temperature of 20° C over 10 orbits to allow the temperature to stabilise and oscillate about a realistic temperature mean. Sources of radiation within the simulated environment include the Sun, Earth albedo and Infrared (IR) re-emission by the planet. A number of mission scenarios were investigated including operational and non-operational modes during nominal, cold, and solstice* cases. The results provide an in orbit minimum and maximum temperature range for the SiPM array, CPLD and MSP430. It also verified the high thermal insulation of the scintillation crystal with respect to the rest of the instrument.

Table 4:

Results of the thermal simulations of GMOD for operational & non-operational modes under a variety of conditions, with a 10°C simulation uncertainty applied (i.e. 10°C added to the predicted maximum, subtracted from the predicted minimum). Highlights indicate temperature extremes experienced across all scenarios.

3.3.3

Test Levels and Set-Up

Through simulation, the maximum and minimum temperature extremes experienced while in orbit was found to be +64.6°C and -20.9°C respectively. For the purposes of qualification testing, an additional ±10°C qualification margin^{31, 32} was added to these temperatures. The minimum qualification temperature was therefore defined as -31°C while the maximum was defined as +75°C. The vacuum chamber would be pumped down to a target of 10^–6mBar with thermal cycling permitted once 10^–4mBar had been achieved. Once the Temperature Reference Point (TRP) had surpassed the qualification threshold the instrument was then set to dwell at the qualification temperature for a duration of 60 minutes. The TRP is a physical location on the IUT chosen as a simplified representation of the unit temperature⁴³ and is used as a reference when determining if and for how long the IUT had surpassed the qualification thresholds. The dwell period is allocated to allow sufficient time for thermal conduction to stabilise throughout the IUT while at the qualification temperature.

Following vibration testing, the disassembled GMOD components were cleaned using IPA and lint free wipes before the PT100 sensor and thermocouples were mounted to the GMOD hardware. These would be used for monitoring the thermal behaviour of the components chosen based on the thermal simulations (section 3.3.2). The simulations suggested that the crystal would be heavily thermally insulated within the detector assembly; for this reason, it was of interest to monitor its behaviour. A type T thermocouple was installed on the crystal housing while a custom endcap was machined with a channel to allow the thermocouple cable to be routed outside once sealed. The PTFE venting channel was enlarged to allow enough room for the cable cross section to pass through. The spacer and endcap are representational only, with minimal modifications to support the testing. The mounting of the crystal thermocouple is shown in Figure 8C and D which shows the PTFE spacers around the scintillator and the raised surface caused by the thermocouple which was mounted using aluminium tape. The rest of the type T thermocouples were mounted onto the MB hardware, one on top of the CoolRunnerII CPLD and the other ontop of the MSP430 microcontroller packages. A type T thermocouple to be used as the TRP was also mounted to the MB PCB on a region in the center of the board void of components or traces (as seen in Figure 8A). This location was chosen such that controlling the temperature at this point would effectively drive the temperature of instrument as a whole and could be used to infer it’s mean temperature. This installation was conducted by CSF operators using aluminium tape, while layering a strip of kapton tape atop to attempt to match the emissivity properties of the PCB.

Figure 8:

GMOD within the TVAC chamber. A) The placement of thermocouples on the MSP430, CPLD and the TRP with kapton tape overlay. B) The placement of GMOD in the TVAC, the throiated welding rods, PT100 mounting and the crystal thermocouple cable. C) The crystal thermocouple mounted on the crystal aluminium container surface. D) The crystal within the detector assembly housing supported with PTFE spacers.

Table 5:

TVAC test levels and durations, with phase overview and durations spent beyond qualification thresholds below. * Note: for cold cycle durations in Time Over Tq, T_QMin is measured as the lower bandwidth threshold of <-30°C, 1°C greater than the predefined T_QMin.

An additional PT100 sensor was mounted on the aluminium housing on the -Y face (with respect to the CRF), kapton tape was not used in this case as the emissivity of the aluminium tape better matched that of the aluminium housing. GMOD was then installed within the S³ TVAC chamber (Figure 8); thermocouple cables were routed through the chamber through a feed through block, while GMOD power and data EGSE were connected using the custom vacuum compatible interface PCB via a D-sub 37 cable. GMOD was inverted within the chamber so the aluminium housing interfaced with the thermal plate, with the addition of aluminium tape to secure good thermal contact. The thoriated welding rods were placed beside the detector assembly along its longest axis as a source of gamma-ray emission for the operational cycle spectral measurements. The PT100 placement, thoriated welding rods and crystal thermocouple sensor cable can be seen in Figure 8B. Externally, the thermocouples were routed to the CSF data acquisition hardware, while the GMOD EGSE connected to the external interface board, providing power, command and data readout via the custom UI, discussed in Section 3.1.

3.3.4

Thermal-Vacuum Test Operation & Results

The TVAC was pumped down from ambient pressure to 10^–6mbar; after reaching a minimum of 10^–4mbar, thermal cycling began with a non-operational thermal cycle later followed by three operational* cycles. The TVAC under vacuum would manage these cycles by varying the temperature at a rate of 2°C/min (via the thermal interface plate) from ambient temperature to +75°C with a dwell at that temperature until the instrument reached thermal equilibrium (originally 1 hour duration). The cycle would then proceed to the cold cycle where the thermal plate would drop in temperature below ambient to -31°C with another dwell, as before to ensure thermal equilibrium. A final half cycle (from ambient temperature to hot and returning to ambient) was done to allow the detector to return to ambient from a hot cycle phase to prevent any condensation or residual moisture build up within the chamber after repressurisation and to discourage dendritic⁴⁴ growth on the SIPHRA silicon die. The TVAC thermal and pressure measurements are shown in Figure 9.

To pass the TVAC test campaign, a series of pass/fail criteria were defined. Those most relevant are listed below. The TVAC testing would be passed if:

During the non-operational cycle, as expected the crystal was found to be very well thermally isolated from the rest of the instrument, with significant lag between the crystal temperature and the other sensor measurements (Figure 9, Phases I & II). The first, non-operational cycle reached its dwell and qualification criteria without incident, after which GMOD was powered up and a serial connection was established while rising from the nonoperational cold cycle, into the operational phases (Phases III to IX). A successful health check was performed around 0°C confirming the instrument had survived the non-operational cycle. GMOD was then commanded to enter operational mode and begin measuring a spectrum leading into the operational cycling stage. The first operational phase (Phase III) demonstrated the crystals extreme thermal insulation; at one point the crystal showed a temperature differential of >70°C between the crystal and the rest of the instrument (referencing the TRP). An unforeseen effect of this was that the large differential allowed the crystal to act as a heat sink, drawing heat through conduction from the rest of the instrument. While the level of thermal insulation of the crystal was predicted, this behaviour was not. This became problematic during the initial hot dwell of Phase III. Just before the TRP reached +75°C, the qualification temperature, it was drawn down below this by the crystal’s heat sink behaviour, below that required to achieve the predefined qualification level. This is seen in Figure 9 just beneath the +75°C mark in Phase III. The CSF operators attempted to mitigate the effects of this by adapting the scripted TVAC control profile resulting in an increased dwell duration from 1 to 2 hours in duration. Despite this, the crystal’s behaviour resulted in multiple dwell stabilisation infractions, compounding the matter, it was impossible to manually adjust the thermal plate setpoint for the remaining two cycles (Phases V - VIII) as it would be performed through automation, overnight and without an attending operator. This would cause the TRP to hover just above the cold dwell threshold qualification of -31°C for all operational cycles (Phase IV, VI, VIII) and exceed at some points the upper limit hot dwell threshold of +78°C (Phases III, V, VII). Regardless, GMOD continued to function as expected without any impact or impedance to performance. An NCR detailing the non conformance was produced outlining the thermal behaviour of the crystal, but stating that there was sufficient confidence that the instrument was properly stressed for qualification and that the objectives of the test were met. This is due in part to the close proximity of the cold dwell qualification temperature to that which was actually reached, a difference within just 1°C. Furthermore, it was deemed that the test levels were far more demanding than those defined in the ECSS standards. Additionally GMOD is yet to undergo system level qualification testing, while integrated with the rest of the EIRSAT-1 payload, which will provide further confidence in the qualification. Upon completion of the fourth and final cycle of the TVAC campaign (Phases VII to VIII), the instrument was returned to ambient from a hot cycle (Phase IX). The instrument was finally removed and disassembled for visual inspection. All thermocouples were removed and components were inspected and photographed. As no observable damage was discovered and GMOD successfully generated a spectrum, all pass/fail criteria were passed successfully. The duration achieved at qualification levels for each phase during the TVAC campaign are shown in Table 5.

Figure 9:

TVAC cycling profile with dwell duration for qualification ranges above in hours. Measurements for the MSP430 and CPLD overlap almost perfectly with the TRP measurements with a maximum difference of ~1.4°C. For the sake of clarity they have been omitted here. The durations in hours above may be also referenced in Table 5 where T_QMin is taken as the lower bandwidth threshold of -30°C.

SiPMs operate above the so-called breakdown voltage, in the case of GMOD this is a negatively applied bias voltage. The overvoltage, the excess voltage above breakdown, defines the internal amplification (gain) of the SiPM. The breakdown voltage is known to depend on temperature. The temperature dependence of SiPM breakdown voltage and the resulting effects on their characteristics are well known⁴⁵ and were expected to be encountered during the testing. For J-Series SiPMs,⁴⁶ the breakdown voltage is known to vary by 21.5mV/°C. In brief, SiPM gain is defined as the amount of charge released per single photon absorption and is a function of the SiPM overvoltage. Since breakdown voltage is temperature dependent, for a fixed applied bias, any variations in temperature will proportionally vary the overvoltage, and by consequence, the SiPM gain. Thus for a scintillation signal produced by a gamma-ray photon (of a given energy) the amount of charge released by the SiPM array will be either increased or decreased. From a high level perspective, for a fixed applied bias voltage and variable temperature, this will result in a shifting of spectral features and a smearing of any long duration spectral integration when performing pulse height spectroscopy. Another effect is a strong influence of temperature on the SiPM dark current and noise, also functions of temperature. SIPHRA readout of the SiPM array is triggered when the current signal from the detector surpasses a fixed trigger threshold level. Normally this threshold is set sufficiently above the thermally generated dark current signal, such that any fluctuations contributed by SiPM noise do not trigger SIPHRA readout. In the event that the average dark current surpasses the trigger threshold due to the strong influence of temperature, SIPHRA will enter a state where it is constantly triggering on dark current signal at its maximum rate, saturating the readout bandwidth with noise events. This can be seen when SIPHRAs output is dominated by successive triggers with low Analogue to Digital Converter (ADC) value.

While in orbit, GMOD would counter these effects by compensation using temperature reference measurements and adjusting accordingly the bias voltage or SIPHRA configuration to maintain consistent performance. However, due to the GMOD firmware being in an early stage of development, these compensation mechanisms were not implemented for the TVAC testing, thus the thermal effects and their impacts are shown in Figure 10. These effects include temperature induced variations on SiPM gain and dark current fluctuation.

Figure 10:

The operational cycle spectral measurements in time resolved format obtained during TVAC cycling. Each slice along the horizontal is an individual spectrum, measured over a given time bin interval. The effects of temperature induced gain variation (as described in Section 3.3.4) are apparent, however the required spectral features remain clearly distinguishable, as seen in Figure 4 & 7. Each phase of the testing has also be labelled as in Figure 9, with the SiPM PT100 data included with a calibration applied for temperature reference.

To examine the detector performance as a function of temperature without smearing from temperature induced gain shifting, the plot is arranged in time resolved format binned into 5 minute intervals (along the y-axis), chronologically from the top down. The shade intensity represents counts per pixel for each ADC bin of spectral data (along the x-axis), in other words, each slice along the horizontal is a individual spectrum at a given time interval, where the most prominent peaks generated by the ²³²Th source can be distinguished by their colour intensity without being smeared or obscured by dark count saturation. A PT100 is present on the underside of the SiPM array PCB which is digitised and read out by SIPHRA upon trigger (as discussed in Section 2). These measurements are also included for reference with a temperature calibration, where again, the shade intensity represents counts per pixel for each ADC bin of PT100 data. From Figure 10, it can be seen that the prominent spectral features (the 238keV line in particular) are shifted down in ADC channels with increase in temperature and up in ADC channels with decrease in temperature. However this continuous shifting of the features is broken by voids across all but the lowest ADC channels, during the rise up to and descent from the hot dwell. At these times the SiPM array has reached such a temperature, that the intensity of the dark current has exceeded the fixed trigger threshold, producing a spectrum saturated with noise events. The instrument then returns to normal operation when the temperature has reduced below ~40–50°C, showing standard readout as the dark current level descends below the trigger threshold. It should be noted, there are a number of ways to mitigate the effects of these behaviours, which unfortunately could not be incorporated into the MSP430 firmware at the time of these tests. This includes bias voltage compensation (to maintain a constant overvoltage) as well as adjustment of the trigger threshold. This final method, was actually demonstrated during the final half cycle (Phase IX) towards the end of the TVAC testing with some success. Figure 10 shows the spectrum during the last hours of the test to be interrupted by saturation only briefly, while the trace of the 232keV peak is just clear enough to be seen consistently. That said this is impractical and quite difficult to perform, particularly if intended to be automated while in flight. Certainly this may need to be adjusted during the mission lifetime (for example, due to cumulative irradiation effects on the SiPMs²¹), however SIPHRA is capable of subtracting the average current before trigger, thus any increase in dark current may be compensated without need of adjusting the threshold level. In the context of the limitations imposed without this being available for the campaign, the instrument performed as expected and was considered to have passed the predefined pass/fail criteria concluding the TVAC campaign.